UBC Faculty Research and Publications

Reductions in intestinal Clostridiales precede the development of nosocomial Clostridium difficile infection Vincent, Caroline; Stephens, David A; Loo, Vivian G; Edens, Thaddeus J; Behr, Marcel A; Dewar, Ken; Manges, Amee R Jun 28, 2013

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


52383-40168_2013_Article_19.pdf [ 587.58kB ]
JSON: 52383-1.0223877.json
JSON-LD: 52383-1.0223877-ld.json
RDF/XML (Pretty): 52383-1.0223877-rdf.xml
RDF/JSON: 52383-1.0223877-rdf.json
Turtle: 52383-1.0223877-turtle.txt
N-Triples: 52383-1.0223877-rdf-ntriples.txt
Original Record: 52383-1.0223877-source.json
Full Text

Full Text

RESEARCH Open AccessReductions in intestinal Clostridiales precede thedevelopment of nosocomial Clostridium difficileinfectionCaroline Vincent1,2, David A Stephens3, Vivian G Loo4, Thaddeus J Edens5, Marcel A Behr1,4, Ken Dewar2,4,6and Amee R Manges7*AbstractBackground: Antimicrobial use is thought to suppress the intestinal microbiota, thereby impairing colonizationresistance and allowing Clostridium difficile to infect the gut. Additional risk factors such as proton-pump inhibitorsmay also alter the intestinal microbiota and predispose patients to Clostridium difficile infection (CDI). Thiscomparative metagenomic study investigates the relationship between epidemiologic exposures, intestinal bacterialpopulations and subsequent development of CDI in hospitalized patients. We performed a nested case–controlstudy including 25 CDI cases and 25 matched controls. Fecal specimens collected prior to disease onset wereevaluated by 16S rRNA gene amplification and pyrosequencing to determine the composition of the intestinalmicrobiota during the at-risk period.Results: The diversity of the intestinal microbiota was significantly reduced prior to an episode of CDI. Sequencescorresponding to the phylum Bacteroidetes and to the families Bacteroidaceae and Clostridiales Incertae Sedis XIwere depleted in CDI patients compared to controls, whereas sequences corresponding to the familyEnterococcaceae were enriched. In multivariable analyses, cephalosporin and fluoroquinolone use, as well as adecrease in the abundance of Clostridiales Incertae Sedis XI were significantly and independently associated withCDI development.Conclusions: This study shows that a reduction in the abundance of a specific bacterial family - ClostridialesIncertae Sedis XI - is associated with risk of nosocomial CDI and may represent a target for novel strategies toprevent this life-threatening infection.Keywords: Intestinal microbiota, Clostridium difficile infection, 16S rRNA gene sequencing, Clostridiales IncertaeSedis XIBackgroundClostridium difficile infection (CDI) is the leading causeof nosocomial diarrhea. The incidence and severity ofCDI have been rising over the last decade and outbreakscontinue to occur across the globe [1]. The changingepidemiology has been linked in part to the emergenceof hypervirulent strains of C. difficile that are resistant tofluoroquinolones [2]. During the major North Americanoutbreak of 2003 to 2005, the proportion of complicatedCDI cases requiring colectomy rose to 18% and fatalityrates reached 25% [3,4]. Recognized risk factors for CDIinclude advanced age, severe underlying illness, previoushospitalization, prolonged hospital stay, and most impor-tantly, exposure to antimicrobials [5]. Broad-spectrumantimicrobial agents are presumed to disrupt the indigen-ous intestinal microbiota, thereby impairing colonizationresistance and allowing the establishment and prolife-ration of C. difficile in the gut. Although nearly all classesof antimicrobial agents have been associated with CDI,clindamycin, penicillins, cephalosporins, and more re-cently fluoroquinolones seem to pose the greatest risk[5-7].* Correspondence: amee.manges@ubc.ca7School of Population and Public Health, University of British Columbia, 2206East Mall, Vancouver, British Columbia V6T 1Z3, CanadaFull list of author information is available at the end of the article© 2013 Vincent et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the CreativeCommons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, andreproduction in any medium, provided the original work is properly cited.Vincent et al. Microbiome 2013, 1:18http://www.microbiomejournal.com/content/1/1/18Other medications besides antimicrobials may alsoalter the intestinal microbiota and predispose patients toCDI. Gastric-acid suppressive agents like proton-pumpinhibitors (PPIs), may act synergistically with antimicro-bial agents to disrupt the intestinal microbiota and con-tribute to CDI development [8]. Epidemiologic evidencehas demonstrated an increased risk of nosocomial CDIin patients receiving PPI therapy, often concurrentlywith antimicrobial agents [9,10]. C. difficile-induced in-flammation is another factor that may, in conjunctionwith antimicrobial use, affect the integrity of the intestinalmicrobiota. Research based on mouse colitis models sug-gests that intestinal inflammation elicited during coloni-zation by enteric pathogens such as Salmonella and C.difficile suppresses the indigenous microbiota, allowingthese invaders to grow unimpeded [11].The objective of this study was to examine the com-plex relationships between epidemiologic exposures, intes-tinal bacterial populations, and subsequent developmentof CDI in hospitalized patients. In a previous investigation,we used a microarray with a limited set of 16S rRNAprobes to contrast the composition of the fecal microbiotabetween patients who later developed CDI (cases) andhospitalized controls. In this earlier study, Firmicutes andBacteroidetes were found to be significantly and inde-pendently associated with CDI development [12]. In orderto validate and expand these initial results, we re-assessedthese valuable pre-disease fecal samples by implementinggold-standard 16S rRNA gene sequencing to obtain acomprehensive survey of the bacterial taxa that are pre-sent in the intestinal tract of patients, and by employingstatistical approaches to appropriately deal with patients’complex exposure histories and the high-dimensional na-ture of the sequencing data. As the composition of the in-testinal microbiota is the unifying theme of this study, wealso adjusted our target epidemiologic exposure windowto focus only on medications received prior to stool col-lection in each patient. We specifically examined (i) pro-files of intestinal microbiota diversity across patients,(ii) differences in the pre-disease composition of the intes-tinal microbiota between CDI cases and control patients,(iii) the association between intestinal bacterial popu-lations and risk of CDI after adjusting for exposure toepidemiologic factors, and (iv) the relationship bet-ween epidemiologic exposures and intestinal micro-biota composition. We report that distinctive featuresof the intestinal microbiota are associated with CDIrisk in hospitalized patients.MethodsStudy design and subjectsBetween September 2006 and May 2007, a total of 599hospitalized patients were enrolled in a prospective co-hort study at the Royal Victoria Hospital, Montréal. Adetailed description of the cohort study is available inLoo et al. [13]. During the study period, 31 patients ex-perienced one or more defined episodes of CDI. Fecalspecimens collected before the onset of the first CDIepisode (if multiple occurred) were available for 25 ofthese patients (cases), and 25 matched controls were se-lected for inclusion in a nested case–control study. Casepatients were matched to controls based on sex, age (± 5years) and date of hospitalization (± 2 months). For thisstudy, a single rectal swab was obtained from each studysubject within 7 days of admission to the hospital. Aquestionnaire was administered to all study patients andcollected information concerning demographics, reasonfor admission, date of admission and discharge, previoushospitalization, underlying disease severity (based onCharlson index), CDI diagnosis, and use of variousmedications in the 8 weeks prior to hospital admis-sion and during hospitalization. Detailed medicationexposures for all patients included in the study areprovided in Additional file 1: Table S1. Informationon nasogastric intubation, aminoglycoside and metro-nidazole use was collected by the larger study, butthese variables were excluded from further analysesas exposure was unknown for a large number of pa-tients. We included exposure to intravenous van-comycin in our analyses, as evidence suggests thatsubstantial amounts of the drug can be excreted inthe bowel and may therefore affect the intestinalmicrobiota [14]. All participants provided informedwritten consent. The human subjects’ protocols forthe cohort and case–control studies were approvedby the Royal Victoria Hospital Internal Review Boardas well as the McGill University Institutional ReviewBoard (BMB 05–014).DefinitionsCDI was defined by the larger cohort study [13] as fol-lows: (i) the presence of diarrhea and a positive C. diffi-cile cytotoxin assay or toxigenic culture, (ii) the presenceof diarrhea without an alternate explanation and anendoscopic diagnosis of pseudomembranes, or (iii) apathological diagnosis of CDI. Diarrhea was defined asthree loose stools within 24 hours for one or more days.Toxigenic C. difficile culture was performed accordingto standard procedures [13].Fecal specimen processingFecal DNA was isolated with use of the DNA IQ System(Promega Corporation, Madison, WI, USA) and subjec-ted to whole-genome amplification using the illustraGenomiPhi V2 DNA Amplification Kit (GE HealthcareBio-Sciences Corporation, Piscataway, NJ, USA). Whole-genome amplification was necessary because of limitedfecal material available from study subjects and toVincent et al. Microbiome 2013, 1:18 Page 2 of 11http://www.microbiomejournal.com/content/1/1/18ensure sufficient DNA quantities for subsequent steps.The GenomiPhi kit was previously shown to generatethe least amount of bias compared to other DNA ampli-fication methods [15]. The amplified DNA was purifiedwith the PureLink PCR Purification Kit (Life Technolo-gies Inc., Burlington, ON, Canada).16S rRNA gene amplification and sequencing16S rRNA gene amplification was performed as de-scribed in the Human Microbiome Project Provisional16S 454 Protocol [16]. Pyrosequencing was performed atthe McGill University and Génome Québec InnovationCentre using Roche/454 GS-FLX Titanium technology.Bioinformatic analysisThe open-source software mothur [17] was used to pro-cess sequences from 16S rRNA gene libraries. Sequencescorresponding to V1-V3 and V3-V5 were binned ac-cording to primer sequence and analyzed separately.Reads containing ambiguous bases (Ns), homopolymerruns greater than 8 bases, inexact match to the MID tag,or more than two differences from the primer sequencewere excluded from the dataset. Remaining sequenceswere aligned against mothur’s Silva reference databaseusing the NAST algorithm [18]. Potential chimeric se-quences were detected with mothur’s implementation ofthe ChimeraSlayer tool [19] and removed accordingly.Rare sequence variants that likely arose from pyrose-quencing errors were merged with their more abundantparent sequence using a single-linkage pre-clustering al-gorithm [17].To determine the proportion of sequences corres-ponding to C. difficile in patient samples, the entireset of high-quality reads from V1-V3 and V3-V5 wereBLASTed against a database of 42 annotated 16S rRNAgenes representing 4 finished C. difficile genomes.BLAST hits with ≥99% identity and ≥99% coverage wereconsidered to be C. difficile.In all other analyses, we controlled for differences insequencing depth by normalizing the number of high-quality reads obtained for each sample. For taxonomicanalyses, sequences were annotated using phylum andfamily-level assignments with the Bayesian classifierimplemented by the Ribosomal Database Project [20]. Aminimum confidence threshold of 80% was required foreach assignment. For diversity analyses, sequences weregrouped into species-level operational taxonomic units(OTUs) using the average neighbor clustering algorithm[21]. OTUs were defined as groups of 16S sequencessharing at least 97% pairwise identity. The heatmap withhierarchical clustering was generated with R [22]. Prin-cipal coordinate analysis (PCoA) was performed withmothur [17].Statistical analysisNormalized sequence counts by bacterial phylum andfamily were log-transformed, in order to minimize un-due influence from extreme values. In statistical modelsincluding epidemiologic variables, we considered thoseexposures that occurred in the 8 weeks before, as well asduring hospitalization, until the date of stool collection.In univariate analyses, we first used a penalized least-squares regression approach (LASSO) to select im-portant predictor variables [23]. The association betweenintestinal bacterial taxa and CDI development was eva-luated by logistic regression. The association betweenepidemiologic exposures and intestinal microbiota com-position was assessed by Poisson regression. An inter-action term was included in the latter model to accountfor the effect of disease status on intestinal bacterialpopulations. Multivariable logistic regression was usedto identify which bacterial taxa remained independentlyassociated with CDI development after adjusting for theeffects of epidemiologic exposures. Selected variablesincluded medications that were administered to atleast 8 out of 50 patients, as well as bacterial taxathat were associated with CDI in the univariate ana-lysis. Multiscale bootstrapping and analysis of mole-cular variance (AMOVA) tests were performed withpvclust and mothur, respectively [17,24]. All other sta-tistical analyses were performed with Stata (version 11,StataCorp) or R [22].ResultsSubject characteristicsTwenty-five patients with CDI and 25 age- and sex-matched controls were retrospectively sampled fromsubjects enrolled in a large cohort study at the RoyalVictoria Hospital, Montréal. Characteristics of the studypatients are reported in Table 1. Case and control pa-tients had similar risk of exposure to C. difficile; theduration of hospitalization (based on the time from ad-mission until CDI diagnosis for case patients or untildischarge for control patients) was not significantly dif-ferent between the groups (P = 0.26, by Mann–WhitneyU-test). Among CDI cases, three patients experiencedmore than one defined episode of CDI during the studyperiod. A single fecal specimen was obtained from eachpatient shortly after hospital admission, but before theonset of the first CDI episode (if multiple occurred) inthe cases (during the at-risk period). Among case pa-tients, the median interval of time between stool collec-tion and CDI diagnosis was 5 days.16S rRNA gene sequencingFecal specimens (n = 50; one per subject) were evaluatedby 16S rRNA gene amplification and pyrosequencing todetermine the composition of the intestinal microbiota.Vincent et al. Microbiome 2013, 1:18 Page 3 of 11http://www.microbiomejournal.com/content/1/1/18A total of 4.1 × 105 high-quality reads (range 2,676-31,641 per subject) from two segments (V1-V3 andV3-V5) of the 16S rRNA gene were analyzed. The ta-xonomic profiles generated from the amplification ofV1-V3 and V3-V5 were in accordance, with a medianPearson correlation coefficient of 0.90. The majority ofsampled sequences corresponded to Firmicutes (51%and 55%), Bacteroidetes (34% and 30%) and Proteobac-teria (8% and 11%) (percentage of V1-V3 and V3-V5 se-quence sets, respectively).Based on toxigenic C. difficile culture assays per-formed by the larger cohort study [13], six patients werefound to be culture-positive on the same date as thefecal specimen used for sequencing was collected. Ofthese, four patients went on to develop CDI (cases), andtwo were asymptomatically colonized (controls). In fiveof these six culture-positive patients, we detected se-quences corresponding to C. difficile; sequences couldnot be detected in one of the two asymptomatically colo-nized controls. There were two instances where C. difficileV1-V3 or V3-V5 sequences comprised >1% of the se-quence data, and both of these patients showed clinicalmanifestation of CDI within the subsequent two days.Intestinal microbiota diversityAfter normalizing for read counts across samples we ob-served variability in intestinal biodiversity across pa-tients. Patient samples exhibited a Shannon Index valuerange of 0.2 to 3.9 according to V1-V3 data and a rangeof 0.3 to 4.2 according to V3-V5 data (Figure 1). Re-duced biodiversity was significantly related to incipientCDI. Based on V1-V3 data, all 8 patients (7 patientsbased on V3-V5 data) with the lowest degree of mi-crobial diversity developed CDI, as did 16 (14 based onV3-V5 data) of the 20 patients with the least diversemicrobiota (Figure 1). Sequence abundances by bacterialfamily across patients are presented as a heatmap withTable 1 Characteristics of study patientsVariable CDI cases(n = 25)Controls(n = 25)Age, mean years ± SD 70 ± 12.8 69 ± 12.5Male sex 12 (48) 12 (48)Charlson comorbidity index, median score (IQR) 1 (1–3) 2 (1–3)Duration of hospitalizationa, median days (IQR) 7 (4–28) 11 (8–17)Hospitalization in past 12 months 19 (76) 15 (60)Reason for hospital admissionCardiac problem 9 (36) 10 (40)Gastrointestinal problem 10 (40) 6 (24)Pulmonary problem 2 (8) 3 (12)Renal disease 2 (8) 1 (4)Otherb 2 (8) 5 (20)Medication usecH2 blocker 7 (29) 4 (16)Nonsteroidal anti-inflammatory drug 15 (65) 12 (48)Proton-pump inhibitor 11 (46) 14 (56)Steroid 3 (13) 1 (4)Any antimicrobial agent 21 (91) 13 (52)Cephalosporind 10 (42) 5 (20)Fluoroquinolonee 9 (38) 4 (16)Macrolide 0 (0) 1 (4)Penicillin 0 (0) 1 (4)Penicillin with β-lactamase inhibitor 7 (30) 3 (12)Vancomycinf 5 (22) 3 (12)Data are number (%) of subjects unless otherwise specified. CDI Clostridiumdifficile infection, SD standard deviation, IQR interquartile range. aDuration untilCDI diagnosis for case patients or duration until discharge for control subjects.bOther reasons include acquired immunodeficiency syndrome, cancer, breastsurgery, anemia, osteomyelitis, neurological or rheumatological problems.cDefined as use within 8 weeks before or during hospitalization, until stoolcollection. Information on steroid, nonsteroidal anti-inflammatory drug, anyantimicrobial agent, penicillin with β-lactamase inhibitor and vancomycin usewas available for 23 out of 25 case patients; information on other medicationswas available for 24 out of 25 case patients. None of the patients wereexposed to probiotics, carbapenem, clindamycin, gatifloxacin, levofloxacin,linezolid, tetracycline or trimethoprim-sulfamethoxazole prior to stoolcollection. dIncludes exposure to first-, second-, and third-generationcephalosporins. eIncludes exposure to ciprofloxacin and moxifloxacin.fIntravenous administration.V1-V3 V3-V5P <0.005 P <0.05Figure 1 Diversity of the intestinal microbiota acrossClostridium difficile infection (CDI) cases and control subjects.The 16S rRNA gene sequences were clustered into operationaltaxonomic units (OTUs) defined by ≥97% nucleotide sequenceidentity. Case patients (n = 25) are colored in red and controlpatients (n = 25) are colored in blue. Patients with the lowestdegree of intestinal biodiversity (n = 6; all of these patients arecases) are shown with open circles. Results are presented for bothV1-V3 and V3-V5 sequence sets. Horizontal lines represent themedian and interquartile range. P values were determined byMann–Whitney U-test.Vincent et al. Microbiome 2013, 1:18 Page 4 of 11http://www.microbiomejournal.com/content/1/1/18hierarchical clustering in Figure 2. There was substantialinter-individual variation in the composition of the fecalmicrobiota. Patient samples were divided in two mainprofile clusters, A and B, with 93% and 89% support, re-spectively, by multiscale bootstrapping with 100,000 rep-licates. Cluster A contained a subset of 8 CDI cases,while cluster B contained all 25 controls unevenly mixedwith 17 cases.Differences in intestinal community membershipacross patients were assessed by PCoA of Jaccard dis-tances (Figure 3). In both V1-V3 and V3-V5 sequencesets, the large majority of intestinal samples from casesand controls clustered separately in the PCoA plot(P <0.001 for both V1-V3 and V3-V5 sequence sets, byAMOVA), although the overall magnitude of variationexplained by principal coordinate 1 was modest (8.1%for V1-V3 and 10.1% for V3-V5). Samples from case pa-tients also displayed a greater level of heterogeneity intheir community membership, as shown by the largerwithin-group distances in cases compared to controls(Figure 3; P <0.0001 for both V1-V3 and V3-V5 se-quence sets, by Mann–Whitney U-test).Association between intestinal microbiota compositionand Clostridium difficile infection (CDI)We examined the relationship between intestinal bacterialtaxa and subsequent development of CDI. At the phylumlevel, the abundance of Bacteroidetes was significantlylower in cases compared to controls (Figure 4A). Atthe family level, sequences corresponding to Clostri-diales Incertae Sedis XI (Figure 4B) and Bacteroidaceae(Figure 4C) were depleted in patients with CDI, while se-quences assigned to Enterococacceae were enriched(Figure 4D).Association between intestinal microbiota compositionand Clostridium difficile infection (CDI) after adjustmentfor epidemiologic exposuresWe used a multivariable analysis to control for the in-fluence of antimicrobials and other medications on thedistribution of bacterial taxa that are related to CDI devel-opment (Table 2). At the phylum level, cephalosporin andfluoroquinolone use were significantly associated withCDI development, while a decrease in the proportion ofBacteroidetes was of borderline significance. At the familyVeBaFiPrAcFu** * ***EnterococcaceaeBacteroidaceaePrevotellaceaePorphyromonadaceaeLachnospiraceaeClostridialesVeillonellaceae04021044541926147StaphylococcaceaeIncertae Sedis XIBAFigure 2 Intestinal microbiota profiles across Clostridium difficile infection (CDI) cases and control subjects. The heatmap shows theabundance of V1-V3 sequences by bacterial family (rows) across all patients (columns). Sequence counts were normalized in order to obtain anequivalent number of reads for each sample. The dendrogram shows hierarchical clustering (unweighted pair group method with arithmeticmean) of microbial communities using Canberra distance metric. The bar on the top indicates disease status for each patient: cases (n = 25) arein red and controls (n = 25) are in blue. Patients with the lowest degree of intestinal biodiversity (n = 6; all of these patients are cases) aremarked with an asterisk. The staggered bars on the left indicate phylum affiliations: Ba, Bacteroidetes; Fi, Firmicutes; Pr, Proteobacteria; Ac,Actinobacteria; Ve, Verrucomicrobia; Fu, Fusobacteria. Other phyla (Lentisphaerae, Spirochaetes, Synergistetes, Tenericutes, Cyanobacteria andTM7) and reads that were unclassified at the phylum level, which altogether represent ≤6.3% of the reads/patient, are not depicted. Relevantbacterial families are listed on the right of the figure. The color gradient is proportional to the logarithm of sequence abundance from 0 to 1,044reads, as indicated by the scale.Vincent et al. Microbiome 2013, 1:18 Page 5 of 11http://www.microbiomejournal.com/content/1/1/18-0.4-0.3-0.2- -0.25 0 0.25PC2 (5.7%)PC1 (10.1%)-0.3-0.2- -0.25 0 0.25 0.5PC2 (5.6%)PC1 (8.1%)A V1-V3 sequence set V3-V5 sequence setBFigure 3 Intestinal community clustering of Clostridium difficile infection (CDI) cases and control subjects based on principalcoordinate analysis (PCoA). Results are presented for (A) V1-V3 and (B) V3-V5 sequence sets. Case patients (n = 25) are colored in red andcontrol patients (n = 25) are colored in blue. Patients with the lowest degree of intestinal biodiversity (n = 6; all of these patients are cases) areshown with open circles. The percentage of variation explained by each principal coordinate (PC) is indicated on the corresponding axis.P<0.05 P<0.05V1-V3 V3-V5P<0.005 P<0.005V1-V3 V3-V5A BP<0.05 P<0.05V1-V3 V3-V5P<0.05 P<0.005V1-V3 V3-V5C DBacteroidetes Clostridiales Incertae Sedis XIBacteroidaceae EnterococcaceaeFigure 4 Intestinal bacterial taxa exhibiting significant differences in abundance between Clostridium difficile infection (CDI) cases andcontrol subjects. The scatter plots show log-transformed 16S sequence counts for the corresponding bacterial (A) phylum or (B-D) family incases (n = 25) versus controls (n = 25). Results are presented for both V1-V3 and V3-V5 sequence sets. Patients with the lowest degree ofintestinal biodiversity (n = 6; all of these patients are cases) are shown with open circles. Horizontal lines represent the median and interquartilerange. P values were determined by logistic regression.Vincent et al. Microbiome 2013, 1:18 Page 6 of 11http://www.microbiomejournal.com/content/1/1/18level, cephalosporin exposure and a reduction in the fre-quency of Clostridiales Incertae Sedis XI were significantrisk factors for CDI, while fluoroquinolone exposure wasof borderline significance.Association between epidemiologic exposures andintestinal microbiota compositionThe impact of antimicrobials and other medications onthe composition of the intestinal microbiota in patientswas examined. We detected a significant associationbetween exposure to penicillin with β-lactamase inhi-bitor and an increase in the abundance of Firmicutes(Figure 5). No other medications were observed to beassociated with differences in intestinal microbiotacomposition.Assessment of six patients with least diverse intestinalmicrobiotaSix case patients exhibited the lowest degree of microbialdiversity (V1-V3 and V3-V5 Shannon Index value <1.7) ofall study subjects (Figure 1). Although medication infor-mation was missing for one case, there was no indicationthat the other five low diversity cases differed from therest of patients in terms of exposure to antimicrobials orTable 2 Multivariable analysis of epidemiologic exposures and intestinal bacterial taxa related to Clostridium difficileinfection (CDI) developmentVariable Phylum-level analysisV1-V3 sequence set V3-V5 sequence setP valuea Coefficient signb P valuea Coefficient signbBacterial phylumBacteroidetes 0.048 - 0.061 -Shannon diversity 0.187 - 0.455 -Medication usecH2 blocker 0.319 - 0.271 -Nonsteroidal anti-inflammatory drug 0.684 + 0.989 +Proton-pump inhibitor 0.443 - 0.467 -Cephalosporin 0.016 + 0.009 +Fluoroquinolone 0.038 + 0.018 +Penicillin with β-lactamase inhibitor 0.228 + 0.424 +Vancomycind 0.278 + 0.116 +Family-level analysisV1-V3 sequence set V3-V5 sequence setP valuea Coefficient signb P valuea Coefficient signbBacterial familyBacteroidaceae 0.073 - 0.051 -Clostridiales Incertae Sedis XI 0.015 - 0.025 -Enterococcaceae 0.942 + 0.246 +Shannon diversity 0.728 + 0.238 +Medication usecH2 blocker 0.384 - 0.318 -Nonsteroidal anti-inflammatory drug 0.921 + 0.605 +Proton-pump inhibitor 0.674 - 0.558 -Cephalosporin 0.020 + 0.027 +Fluoroquinolone 0.045 + 0.061 +Penicillin with β-lactamase inhibitor 0.692 + 0.850 -Vancomycind 0.423 + 0.193 +aP values were determined by multivariate logistic regression. bA positive sign indicates that 16S sequence abundance (for bacterial taxa) or number of patientsexposed (for medications) was higher among patients with CDI than among controls, while a negative sign indicates that 16S sequence abundance or number ofpatients exposed was lower among patients with CDI. cDefined as use within 8 weeks before or during hospitalization, until stool collection. dIntravenousadministration.Vincent et al. Microbiome 2013, 1:18 Page 7 of 11http://www.microbiomejournal.com/content/1/1/18other medications (see Additional file 1: Table S1; P >0.05for all medications, by Fisher’s exact test). In the heatmap,these six low diversity cases were spread across clusters Aand B and did not share a common taxonomic profile(Figure 2). However, these patients were clearly positionedaway from the cluster of controls in the PCoA plot(Figure 3). The enrichment in Enterococacceae wasobserved in all of the six patients with reduced bio-diversity (Figure 4D).DiscussionExposure to antimicrobials or antimicrobials in conjunc-tion with other medications is thought to alter the intes-tinal microbiota and impair colonization resistance to C.difficile. By obtaining fecal specimens in the at-riskperiod prior to CDI onset, we were able to evaluate theimpact of epidemiologic exposures and intestinal mi-crobiota composition on CDI risk. Not only do ourresults confirm the existence of a compromised gutmicrobiota in CDI patients, but we were able to iden-tify specific epidemiologic and microbiota factors thatare significantly and independently associated withCDI development.Several studies have observed a reduced microbial di-versity in patients with CDI or other diseases, includingirritable bowel syndrome and obesity [25-28]. Our re-sults confirm that low diversity is related to CDI devel-opment. However, this feature was not found to be anindependent predictor of CDI in the multivariable ana-lysis. Reduced diversity may be a non-specific marker ofdisease.The levels of Bacteroidetes and Enterococcaceae weremarkedly altered in patients that were about to experi-ence CDI; however, after adjustment for medication use,these associations were no longer significant. Ferreiraet al. have suggested that Bacteroidetes may conferresistance to infectious colitis by protecting againstpathogen-mediated intestinal inflammation [29]. Intri-guingly, the observed increase in Enterococcaceae ap-peared to be mostly driven by a subset of six cases withthe lowest degree of intestinal diversity. Enterococci areopportunistic microorganisms that can, like C. difficile,exploit the reduced biodiversity of the intestinal ecosys-tem to expand their population. This idea is consistentwith studies showing increased levels of enterococci inthe gut following treatment with extended-spectrumantimicrobial agents [30,31]. In a study by Lawley andcolleagues, antibiotic treatment of mice asymptomati-cally colonized with C. difficile resulted in a dramatic re-duction in intestinal microbial diversity accompanied byan expansion of Escherichia coli and enterococci whichtriggered the overgrowth of C. difficile [32].In the multivariable analysis, cephalosporin andfluoroquinolone exposure, as well as a decrease in theabundance of Clostridiales Incertae Sedis XI were signifi-cantly associated with CDI development. According tocurrent taxonomic lineages, C. difficile (which is part ofthe Peptostreptococaceae family) and the bacterial familyClostridiales Incertae Sedis XI belong to the same order(Clostridiales) [20,33]. Therefore, the depletion of Clos-tridiales Incertae Sedis XI that preceded CDI onset mayindicate an absence of competitive exclusion or othercolonization resistance mechanisms operating in the in-testinal microbiota of these patients. Studies involvinganimal models suggest that competition for similar nu-trient sources or ecological niches mediated by closelyrelated bacterial groups that are already established inthe gut may prevent invasion by pathogenic relativessuch as C. difficile [34]. A randomized clinical trial toevaluate the safety and efficacy of colonization with non-toxigenic C. difficile for the prevention of recurrent CDIis currently underway [35].We have demonstrated an association between the useof penicillin with β-lactamase inhibitor and an increasein the abundance of Firmicutes. Culture-based analysesof the human fecal microbiota have previously shown thatadministration of amoxicillin-clavulanic acid (a penicillinP <0.05 P <0.05V1-V3 V3-V5Figure 5 Exposure to penicillin with β-lactamase inhibitor isassociated with an increase in Firmicutes. The scatter plot showslog-transformed 16S sequence counts for Firmicutes in patients thatwere exposed (n = 10) or unexposed (n = 38) to penicillin withβ-lactamase inhibitor. Results are presented for both V1-V3 andV3-V5 sequence sets. Patients with the lowest degree of intestinalbiodiversity (n = 5) are shown with open circles. Horizontal linesrepresent the median and interquartile range. P values weredetermined by Poisson regression. Note that data on exposure topenicillin with β-lactamase inhibitor was missing in two patients,including one of the patients with the lowest degree of intestinalbiodiversity; these are not depicted.Vincent et al. Microbiome 2013, 1:18 Page 8 of 11http://www.microbiomejournal.com/content/1/1/18with β-lactamase inhibitor) increases the number ofaerobic Gram-positive cocci, most of which belong toFirmicutes [36].In our previous 16S rRNA microarray-based investiga-tion of the same set of samples, we could not establishthat low intestinal microbial diversity is associated withCDI development [12]. In this study, high-resolution se-quencing along with analyses performed at a lowerphylogenetic level allowed us to capture most of thebacterial diversity, and we were able to confirm that re-duced diversity is related to CDI. This study also vali-dates our previous observation that an enrichment ofEnterococcaceae and a depletion of Bacteroidetes orClostridiales Incertae Sedis XI are significantly associ-ated with CDI development [12]. We did observe higherlevels of Firmicutes among our CDI patients, as reportedpreviously, but the association was not statistically sig-nificant in the current sequence-based study (P = 0.09,by logistic regression).Other authors have assessed intestinal microbiota al-terations in patients with an initial or recurrent episodeof CDI [25-27,37]. However, these investigations did notaccount for the influence of antimicrobials and othermedications in the analysis of microbial profiles associ-ated with CDI. Moreover, previous studies have typicallyassessed microbiota composition at the time of CDIdiagnosis, when the results are likely confounded by theeffects of the disease itself (that is, diarrhea and intes-tinal inflammation) and the effects of CDI treatment onthe intestinal ecology. We did observe reduced levels ofthe Bacteroides-Porphyromonas-Prevotella group andincreased levels of facultative anaerobes in patientswith CDI, as reported elsewhere, but we did not finda significant association with members of the ente-robacteria, bifidobacteria or lactobacilli [25,26,37]. Amongour 25 case patients, 3 experienced multiple CDI epi-sodes. We did not observe specific microbiota altera-tions that could distinguish these patients from otherCDI cases and the small number of patients pre-cluded any further analyses. Whether specific micro-biota signatures can predict the eventual developmentof recurrent CDI (as opposed to a single CDI episode) re-mains to be addressed.De La Cochetière et al. investigated the relationshipbetween dominant gut bacterial species and subsequentacquisition of C. difficile in outpatients receiving anti-microbial therapy. Fecal samples obtained prior to theinitiation of antimicrobial treatment were analyzed bytemporal temperature gradient gel electrophoresis andthe resulting microbial profiles could accurately predictthe risk of C. difficile acquisition in these subjects [38].Similarly, our results support the idea that certain pa-tients have an existing predisposition to CDI when theyare admitted to the hospital; their intestinal microbiotamay be less resilient to the effects of antibiotics or morepermissive to the invasion of C. difficile.This study is limited by biases inherent to bacterialDNA extraction (due to differential lysis efficiency),whole-genome amplification and 16S rRNA gene ampli-fication (due to species coverage of the primers and vari-able numbers of 16S rRNA gene copies per genome),which may contribute to under- or over-representationof certain bacterial taxa. Our limited sample size alsomade it difficult to account for variability in microbiotaprofiles due to differences in underlying disease andtreatment histories across patients. Despite these limita-tions, important differences in the abundance of keybacterial taxa were apparent and clearly distinguishedCDI cases from control patients.ConclusionsAlthough the association between antimicrobial use andCDI is well established, specific alterations to the intestinalmicrobiota and how they contribute to disease deve-lopment are poorly described. In this study, we identifiedspecific epidemiologic and microbiota factors that are as-sociated with CDI risk in hospitalized patients. Based onmultivariable analyses, independent risk factors for CDIincluded cephalosporin and fluoroquinolone exposure, aswell as a depletion of Clostridiales Incertae Sedis XI. Thisimportant novel finding may eventually lead to the elabo-ration of targeted microbiota interventions to prevent thedevelopment of CDI in high-risk patients.Additional fileAdditional file 1: Table S1. Detailed medication exposures of studypatients.AbbreviationsCDI: Clostridium difficile infection; OTU: Operational taxonomic unit;PCoA: Principal coordinate analysis; PPI: Proton-pump inhibitor; PCR:Polymerase chain reaction; V1-V3: Variable regions 1 to 3 of the 16Sribosomal RNA gene; V3-V5: Variable regions 3 to 5 of the 16S ribosomalRNA gene.Competing interestsThe authors declare that they have no competing interests.Authors’ contributionsCV carried out the study, participated in data analysis and interpretation, anddrafted the manuscript. DAS assisted with statistical analyses. VGL generouslyprovided patient epidemiologic information and clinical samples. TJEperformed the statistical analyses. MAB critically reviewed the manuscript.KD participated in data analysis and interpretation, and helped to draft themanuscript. ARM conceived and designed the study, participated in dataanalysis and interpretation, and helped to draft the manuscript. All authorsread and approved the final manuscript.AcknowledgementsThe authors would like to thank Dr. Nandini Dendukuri and Ian Schiller fortheir assistance with data management. This work was supported by aCatalyst Grant [CHM-94228 to A.R.M.] and a Doctoral Research Award[GSD-113375 to C.V.] from the Canadian Institutes of Health Research.Vincent et al. Microbiome 2013, 1:18 Page 9 of 11http://www.microbiomejournal.com/content/1/1/18Author details1Department of Microbiology and Immunology, McGill University, 3775University Street, Montréal, Québec H3A 2B4, Canada. 2McGill University andGénome Québec Innovation Centre, 740 Dr. Penfield Avenue, Montréal,Québec H3A 0G1, Canada. 3Department of Mathematics and Statistics, McGillUniversity, 805 Sherbrooke Street West, Montréal, Québec H3A 0B9, Canada.4The Research Institute of the McGill University Health Centre, 2155 GuyStreet, Montréal, Québec H3H 2R9, Canada. 5Devil’s Staircase Consulting, 693Osborne Road East, North Vancouver, British Columbia V7N 1M8, Canada.6Department of Human Genetics, McGill University, 1205 Dr. Penfield Avenue,Montréal, Québec H3A 1B1, Canada. 7School of Population and PublicHealth, University of British Columbia, 2206 East Mall, Vancouver, BritishColumbia V6T 1Z3, Canada.Received: 5 February 2013 Accepted: 21 June 2013Published: 28 June 2013References1. Kuijper EJ, Coignard B, Tull P: Emergence of Clostridium difficile-associateddisease in North America and Europe. Clin Microbiol Infect 2006,12(Suppl 6):2–18.2. Razavi B, Apisarnthanarak A, Mundy LM: Clostridium difficile: emergenceof hypervirulence and fluoroquinolone resistance. Infection 2007,35:300–307.3. Loo VG, Poirier L, Miller MA, Oughton M, Libman MD, Michaud S,Bourgault AM, Nguyen T, Frenette C, Kelly M, Vibien A, Brassard P,Fenn S, Dewar K, Hudson TJ, Horn R, René P, Monczak Y, Dascal A: Apredominantly clonal multi-institutional outbreak of Clostridiumdifficile-associated diarrhea with high morbidity and mortality. N EnglJ Med 2005, 353:2442–2449.4. Pepin J, Valiquette L, Alary ME, Villemure P, Pelletier A, Forget K, Pepin K,Chouinard D: Clostridium difficile-associated diarrhea in a region ofQuebec from 1991 to 2003: a changing pattern of disease severity.Cmaj 2004, 171:466–472.5. Ananthakrishnan AN: Clostridium difficile infection: epidemiology,risk factors and management. Nat Rev Gastroenterol Hepatol 2011,8:17–26.6. Owens RC Jr, Donskey CJ, Gaynes RP, Loo VG, Muto CA: Antimicrobial-associated risk factors for Clostridium difficile infection. Clin Infect Dis2008, 46(Suppl 1):S19–S31.7. Pepin J, Saheb N, Coulombe MA, Alary ME, Corriveau MP, Authier S,Leblanc M, Rivard G, Bettez M, Primeau V, Nguyen M, Jacob CE, Lanthier L:Emergence of fluoroquinolones as the predominant risk factor forClostridium difficile-associated diarrhea: a cohort study during anepidemic in Quebec. Clin Infect Dis 2005, 41:1254–1260.8. Kwok CS, Arthur AK, Anibueze CI, Singh S, Cavallazzi R, Loke YK: Risk ofClostridium difficile infection with acid suppressing drugs and antibiotics:meta-analysis. Am J Gastroenterol 2012, 107:1011–1019.9. Dial S, Alrasadi K, Manoukian C, Huang A, Menzies D: Risk ofClostridium difficile diarrhea among hospital inpatients prescribedproton pump inhibitors: cohort and case–control studies. Cmaj 2004,171:33–38.10. Leonard J, Marshall JK, Moayyedi P: Systematic review of the risk ofenteric infection in patients taking acid suppression. Am J Gastroenterol2007, 102:2047–2056. quiz 2057.11. Stecher B, Hardt WD: Mechanisms controlling pathogen colonization ofthe gut. Curr Opin Microbiol 2011, 14:82–91.12. Manges AR, Labbe A, Loo VG, Atherton JK, Behr MA, Masson L, Tellis PA,Brousseau R: Comparative metagenomic study of alterations to theintestinal microbiota and risk of nosocomial Clostridum difficile-associated disease. J Infect Dis 2010, 202:1877–1884.13. Loo VG, Bourgault AM, Poirier L, Lamothe F, Michaud S, Turgeon N, Toye B,Beaudoin A, Frost EH, Gilca R, Brassard P, Dendukuri N, Béliveau C, OughtonM, Brukner I, Dascal A: Host and pathogen factors for Clostridium difficileinfection and colonization. N Engl J Med 2011, 365:1693–1703.14. Currie BP, Lemos-Filho L: Evidence for biliary excretion of vancomycininto stool during intravenous therapy: potential implications for rectalcolonization with vancomycin-resistant enterococci. Antimicrob AgentsChemother 2004, 48:4427–4429.15. Pinard R, de Winter A, Sarkis GJ, Gerstein MB, Tartaro KR, Plant RN,Egholm M, Rothberg JM, Leamon JH: Assessment of whole genomeamplification-induced bias through high-throughput, massivelyparallel whole genome sequencing. BMC Genomics 2006, 7:216.16. Provisional 16S 454 protocol. [http://www.hmpdacc.org/tools_protocols/tools_protocols.php]17. Schloss PD, Westcott SL, Ryabin T, Hall JR, Hartmann M, Hollister EB,Lesniewski RA, Oakley BB, Parks DH, Robinson CJ, Sahl JW, Stres B,Thallinger GG, Van Horn DJ, Weber CF: Introducing mothur: open-source,platform-independent, community-supported software for describingand comparing microbial communities. Appl Environ Microbiol 2009,75:7537–7541.18. DeSantis TZ Jr, Hugenholtz P, Keller K, Brodie EL, Larsen N, Piceno YM,Phan R, Andersen GL: NAST: a multiple sequence alignment serverfor comparative analysis of 16S rRNA genes. Nucleic Acids Res 2006,34:W394–W399.19. Haas BJ, Gevers D, Earl AM, Feldgarden M, Ward DV, Giannoukos G, Ciulla D,Tabbaa D, Highlander SK, Sodergren E, Methé B, DeSantis TZ, Petrosino JF,Knight R, Birren BW, Human Microbiome Consortium: Chimeric 16S rRNAsequence formation and detection in Sanger and 454-pyrosequencedPCR amplicons. Genome Res 2011, 21:494–504.20. Wang Q, Garrity GM, Tiedje JM, Cole JR: Naive Bayesian classifier forrapid assignment of rRNA sequences into the new bacterial taxonomy.Appl Environ Microbiol 2007, 73:5261–5267.21. Schloss PD, Westcott SL: Assessing and improving methods used inoperational taxonomic unit-based approaches for 16S rRNA genesequence analysis. Appl Environ Microbiol 2011, 77:3219–3226.22. The R project for statistical computing. [http://www.r-project.org/]23. Tibshirani R: Regression shrinkage and selection via the Lasso. J R Statist1996, 58:267–288.24. Suzuki R, Shimodaira H: Pvclust: an R package for assessing theuncertainty in hierarchical clustering. Bioinformatics 2006, 22:1540–1542.25. Hopkins MJ, Sharp R, Macfarlane GT: Age and disease related changesin intestinal bacterial populations assessed by cell culture, 16S rRNAabundance, and community cellular fatty acid profiles. Gut 2001,48:198–205.26. Rea MC, O’Sullivan O, Shanahan F, O’Toole PW, Stanton C, Ross RP, Hill C:Clostridium difficile carriage in elderly subjects and associated changes inthe intestinal microbiota. J Clin Microbiol 2012, 50:867–875.27. Chang JY, Antonopoulos DA, Kalra A, Tonelli A, Khalife WT, Schmidt TM,Young VB: Decreased diversity of the fecal Microbiome in recurrentClostridium difficile-associated diarrhea. J Infect Dis 2008, 197:435–438.28. Gerritsen J, Smidt H, Rijkers GT, de Vos WM: Intestinal microbiota inhuman health and disease: the impact of probiotics. Genes Nutr 2011,6:209–240.29. Ferreira RB, Gill N, Willing BP, Antunes LC, Russell SL, Croxen MA, Finlay BB:The intestinal microbiota plays a role in Salmonella-induced colitisindependent of pathogen colonization. PLoS One 2011, 6:e20338.30. Donskey CJ, Chowdhry TK, Hecker MT, Hoyen CK, Hanrahan JA, Hujer AM,Hutton-Thomas RA, Whalen CC, Bonomo RA, Rice LB: Effect of antibiotictherapy on the density of vancomycin-resistant enterococci in the stoolof colonized patients. N Engl J Med 2000, 343:1925–1932.31. Ubeda C, Taur Y, Jenq RR, Equinda MJ, Son T, Samstein M, Viale A, Socci ND,van den Brink MR, Kamboj M, Pamer EG: Vancomycin-resistantEnterococcus domination of intestinal microbiota is enabled by antibiotictreatment in mice and precedes bloodstream invasion in humans. J ClinInvest 2010, 120:4332–4341.32. Lawley TD, Clare S, Walker AW, Goulding D, Stabler RA, Croucher N,Mastroeni P, Scott P, Raisen C, Mottram L, Fairweather NF, Wren BW,Parkhill J, Dougan G: Antibiotic treatment of Clostridium difficilecarrier mice triggers a supershedder state, spore-mediatedtransmission, and severe disease in immunocompromised hosts.Infect Immun 2009, 77:3661–3669.33. Ludwig W, Schleifer K-H, Whitman WB: Bergey’s Manual of SystematicBacteriology. Volume 3. 2nd edition. New York: Springer Science+BusinessMedia; 2009:1–13.34. Britton RA, Young VB: Interaction between the intestinal microbiota andhost in Clostridium difficile colonization resistance. Trends Microbiol 2012,20:313–319.35. Safety and efficacy study of VP20621 for prevention of recurrent Clostridiumdifficile infection. [http://clinicaltrials.gov/ct2/show/NCT01259726]36. Sullivan A, Edlund C, Nord CE: Effect of antimicrobial agents on theecological balance of human microflora. Lancet Infect Dis 2001, 1:101–114.Vincent et al. Microbiome 2013, 1:18 Page 10 of 11http://www.microbiomejournal.com/content/1/1/1837. Hopkins MJ, Sharp R, Macfarlane GT: Variation in human intestinalmicrobiota with age. Dig Liver Dis 2002, 34(Suppl 2):S12–S18.38. De La Cochetière MF, Durand T, Lalande V, Petit JC, Potel G, Beaugerie L:Effect of antibiotic therapy on human fecal microbiota and therelation to the development of Clostridium difficile. Microb Ecol 2008,56:395–402.doi:10.1186/2049-2618-1-18Cite this article as: Vincent et al.: Reductions in intestinal Clostridialesprecede the development of nosocomial Clostridium difficile infection.Microbiome 2013 1:18.Submit your next manuscript to BioMed Centraland take full advantage of: • Convenient online submission• Thorough peer review• No space constraints or color figure charges• Immediate publication on acceptance• Inclusion in PubMed, CAS, Scopus and Google Scholar• Research which is freely available for redistributionSubmit your manuscript at www.biomedcentral.com/submitVincent et al. Microbiome 2013, 1:18 Page 11 of 11http://www.microbiomejournal.com/content/1/1/18


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items